Davyd Norris from Rational's offices in Australia wrote to report that a few years ago, they had received a request from Malaysia for R1000 spare parts. Greg Bek and Stu Garrow recalled that our Adelaide office had kept a number of old R1000s (I remember autographing one of those machines on a trip long ago), machines that were then stored away and largely forgotten. As it turns out, the wheels of bureaucracy spun on and someone somewhere was still paying for that storage unit. The team tracked down the storage company, located the unit, cut open the lock, and discovered a stash of very well-preserved machines. Sadly, sometime after the IBM acquisition and due diligence was done on all of Rational's assets, a team came in and crushed, melted, recycled, and essentially discarded those machines. Alas, such is the fate of all things; dust to dust, ashes to ashes, transistors to silicon.
Stu and Greg established a company serving as a Rational business partner, and apparently rescued a complete working R1000 400 series, which is now sitting in Greg's garage. As Davyd reported, the list price back then was about one million US, so if any of you are hankering for an Ada machine, there you go.
Joe Marasco and Dave Bernstein continue their dialog with me, and recently wrote with some clarifications and additions. Dave noted that the compiler used for Rational's first bootstrapping work was the ADAS compiler (addressing a sequential subset of Ada; back then, talk of Ada subsets was heresy) from David Luckam's group at Stanford, a group that included not only Howard Larsen and Dave Stevenson (whom I'd mentioned in a previous blog) but also Wolf Pollak. Dave went on to clarify that the model 100 R1000 used a PDP 11/24 as the I/O processor and that the model 200 used a Motorola 68000 microprocessor that provided a DEC Unibus, with the hardware and software work led by Wayne Meretsky.
Continuing with Rose's history, Dave observed that it took about five years for Rose to achieve critical mass. Rose 1.0 was a failure for a number of reasons and was painfully withdrawn from the market (I still have an original shrink-wrapped box 'o software). This was Rational's first non-Ada project, and among other things we had made an architectural decision to use an early OODBMS that in the end proved to be a profoundly bad idea. Jon Hopkin's company, Palladio, provided us with a PC-based Booch and OMT method tool. This acquisition was initially precarious yet was brought back from near-death by Dave and Nick Berens. Dave took over the Rose effort in 1994 and in 1995 development of Rose 2.0 began, with Loren Archer as marketing director and Greg Myers as the Rose Business Unit manager. Rich Reitman (who is now at Adobe) developed a version control strategy for Rose, a decision that helped differentiate Rose in the market and greatly contributed to our development process. Simultaneously, Joe took on the Microsoft/RBU coordination, wherein we bought Visual Test from Microsoft and brought them on as a sponsor of the UML, a partnership that gave the UML considerable momentum. Dave Stevenson, Mats Goethe, Jack Tilford, Tom Wilcox, Howard Larsen, Jim Archer, and Adam Frankl were the members of this team, whose efforts led to the Rational/Microsoft announcement on October 5th, 1996 (at OOPSLA, if I'm not mistaken) of UML tooling within Visual Studio. In the end, the work that the Three Amigos (Jim Rumbaugh, Ivar Jacobson, and me) did culminated in Rational forming an amazing constellation of support (from Microsoft, IBM, HP, and Oracle....who would have ever thought that these companies could come together in agreement?) that pushed the UML and modeling in general over the edge. With Rational at the center and Rose good enough, Rose revenues crossed an important threshold that by 1997 had contributed to a dramatic growth in Rational's valuation such that we we able to acquire a number of other companies, that in turn leading to the Rational Unified Process and the Rational Suites. Throughout all this, revenue from Apex provided the bridge that gave Rose time to find its way; without Apex, Rose could have never flouished.
A bit more on Palladio from Dave. Rose 2.0 was essentially Palladio's Windows-based tool (then called the Object System Designer), converted to use Petal as its intermediate representation (we were all about flower images back then). Dave Stevenson did that work, with Frank Tadman contributing the C++ forward engineering bits and Tom Wilcox doing the C++ reverse engineering bits, leading to what we called roundtrip engineering (adapted from Mike Druke's phrase, roundtrip gestalt problem solving). Our nascent field teams (with Alex Baran, Tom Schultz, and Terry Quatrani aka TQ aka Mom) found that the ability to harvest models from as-built systems was the feature that really pushed Rose 2.0 over the edge of value and acceptance.
Quote of the day: